From b9b12ac92230cc6807623293b93cc22e5eb33684 Mon Sep 17 00:00:00 2001
From: Eric Wong <e@80x24.org>
Date: Fri, 3 Mar 2017 00:55:07 +0000
Subject: repobrowse: consistently set text charset

For everything with relevant content, we'll try to set
UTF-8 charset and reduce duplication when generating
response headers.
---
 lib/PublicInbox/RepoBase.pm       |  9 ++++++++-
 lib/PublicInbox/RepoGitAtom.pm    |  3 +--
 lib/PublicInbox/RepoGitCommit.pm  | 11 +++++------
 lib/PublicInbox/RepoGitDiff.pm    |  9 ++++-----
 lib/PublicInbox/RepoGitLog.pm     |  6 +++---
 lib/PublicInbox/RepoGitPatch.pm   |  4 ++--
 lib/PublicInbox/RepoGitRaw.pm     | 10 +++++-----
 lib/PublicInbox/RepoGitSrc.pm     | 31 ++++++++++++-------------------
 lib/PublicInbox/RepoGitSummary.pm |  2 +-
 lib/PublicInbox/RepoGitTag.pm     |  5 ++---
 t/repobrowse_git_log.t            |  2 +-
 11 files changed, 44 insertions(+), 48 deletions(-)

diff --git a/lib/PublicInbox/RepoBase.pm b/lib/PublicInbox/RepoBase.pm
index 9876cf2a..e600b1df 100644
--- a/lib/PublicInbox/RepoBase.pm
+++ b/lib/PublicInbox/RepoBase.pm
@@ -105,10 +105,17 @@ sub r {
 		# mainly for curl (no-'-L') users:
 		$body = "Redirecting to $redir\n";
 	} else {
-		push @h, qw(Content-Type text/plain);
+		push @h, 'Content-Type', 'text/plain; charset=UTF-8';
 	}
 
 	[ $status, \@h, [ $body ] ]
 }
 
+sub rt {
+	my ($self, $status, $t) = @_;
+	my $res = [ $status, [ 'Content-Type', "text/$t; charset=UTF-8" ] ];
+	$res->[2] = [ $_[3] ] if defined $_[3];
+	$res;
+}
+
 1;
diff --git a/lib/PublicInbox/RepoGitAtom.pm b/lib/PublicInbox/RepoGitAtom.pm
index c30c6184..c2393b38 100644
--- a/lib/PublicInbox/RepoGitAtom.pm
+++ b/lib/PublicInbox/RepoGitAtom.pm
@@ -124,8 +124,7 @@ sub git_atom_cb {
 		my $env = $req->{env};
 		if (!defined $r) {
 			my $git = $req->{-repo}->{git};
-			return [ 400, [ 'Content-Type', 'text/plain' ],
-				[ $git->err ] ];
+			return $self->rt(400, 'plain', $git->err);
 		}
 		$env->{'qspawn.filter'} = git_atom_sed($self, $req);
 		[ 200, [ 'Content-Type', 'application/atom+xml' ] ];
diff --git a/lib/PublicInbox/RepoGitCommit.pm b/lib/PublicInbox/RepoGitCommit.pm
index 8add3006..acac000e 100644
--- a/lib/PublicInbox/RepoGitCommit.pm
+++ b/lib/PublicInbox/RepoGitCommit.pm
@@ -135,19 +135,18 @@ sub call_git_commit { # RepoBase calls this
 	$qsp->psgi_return($env, undef, sub { # parse header
 		my ($r, $bref) = @_;
 		if (!defined $r) {
-			my $errmsg = $git->err;
-			[ 500, [ 'Content-Type', 'text/html' ], [ $errmsg ] ];
+			$self->rt(500, 'plain', $git->err);
 		} elsif ($r == 0) {
-			git_commit_404($req);
+			git_commit_404($self, $req);
 		} else {
 			$env->{'qspawn.filter'} = git_commit_sed($self, $req);
-			[ 200, [ 'Content-Type', 'text/html' ] ];
+			$self->rt(200, 'html');
 		}
 	});
 }
 
 sub git_commit_404 {
-	my ($req) = @_;
+	my ($self, $req) = @_;
 	my $x = 'Missing commit or path';
 	my $pfx = "$req->{relcmd}commit";
 
@@ -156,7 +155,7 @@ sub git_commit_404 {
 	$x .= "<a\nhref=\"$pfx\">$try the latest commit in HEAD</a>\n";
 	$x .= '</pre></body>';
 
-	[404, ['Content-Type', 'text/html'], [ $x ]];
+	$self->rt(404, 'html', $x);
 }
 
 # FIXME: horrifically expensive...
diff --git a/lib/PublicInbox/RepoGitDiff.pm b/lib/PublicInbox/RepoGitDiff.pm
index 8d3b97a0..35bcb2d9 100644
--- a/lib/PublicInbox/RepoGitDiff.pm
+++ b/lib/PublicInbox/RepoGitDiff.pm
@@ -54,15 +54,14 @@ sub call_git_diff {
 	$qsp->psgi_return($env, undef, sub { # parse header
 		my ($r) = @_;
 		if (!defined $r) {
-			[ 500, [ 'Content-Type', 'text/plain' ], [ $git->err ]];
+			$self->rt(500, 'plain', $git->err);
 		} elsif ($r == 0) {
-			[ 200, [ 'Content-Type', 'text/html' ], [
+			$self->rt(200, 'html',
 				delete($req->{dhtml}).
-				'No differences</pre></body></html>' ]
-			]
+				'No differences</pre></body></html>');
 		} else {
 			$env->{'qspawn.filter'} = git_diff_sed($self, $req);
-			[ 200, [ 'Content-Type', 'text/html' ] ];
+			$self->rt(200, 'html');
 		}
 	});
 }
diff --git a/lib/PublicInbox/RepoGitLog.pm b/lib/PublicInbox/RepoGitLog.pm
index 1ad83fc0..73319633 100644
--- a/lib/PublicInbox/RepoGitLog.pm
+++ b/lib/PublicInbox/RepoGitLog.pm
@@ -139,12 +139,12 @@ sub call_git_log {
 	$qsp->psgi_return($env, undef, sub {
 		my ($r) = @_;
 		if (!defined $r) {
-			[ 500, [ 'Content-Type', 'text/html' ], [ $git->err ] ];
+			$self->rt(500, 'html', $git->err);
 		} elsif ($r == 0) {
-			[ 404, [ 'Content-Type', 'text/html' ], [ $git->err ] ];
+			$self->rt(404, 'html', $git->err);
 		} else {
 			$env->{'qspawn.filter'} = git_log_sed($self, $req);
-			[ 200, [ 'Content-Type', 'text/html' ] ];
+			$self->rt(200, 'html');
 		}
 	});
 }
diff --git a/lib/PublicInbox/RepoGitPatch.pm b/lib/PublicInbox/RepoGitPatch.pm
index cd5bf13c..1c9ec45f 100644
--- a/lib/PublicInbox/RepoGitPatch.pm
+++ b/lib/PublicInbox/RepoGitPatch.pm
@@ -31,8 +31,8 @@ sub call_git_patch {
 	my $qsp = PublicInbox::Qspawn->new($cmd);
 	$qsp->psgi_return($env, undef, sub {
 		my ($r) = @_;
-		my $h = ['Content-Type', 'text/plain; charset=UTF-8'];
-		$r ? [ 200, $h ] : [ 500, $h, [ "format-patch error\n" ] ];
+		$r ? $self->rt(200, 'plain') :
+			$self->rt(500, 'plain', "format-patch error\n");
 	});
 }
 
diff --git a/lib/PublicInbox/RepoGitRaw.pm b/lib/PublicInbox/RepoGitRaw.pm
index f02439ad..7d2c0d22 100644
--- a/lib/PublicInbox/RepoGitRaw.pm
+++ b/lib/PublicInbox/RepoGitRaw.pm
@@ -22,10 +22,10 @@ sub call_git_raw {
 	if ($type eq 'blob') {
 		$type = git_blob_mime_type($self, $req, $cat, \$buf, \$left);
 	} elsif ($type eq 'commit' || $type eq 'tag') {
-		$type = 'text/plain';
+		$type = 'text/plain; charset=UTF-8';
 	} elsif ($type eq 'tree') {
 		$git->cat_file_finish($left);
-		return git_tree_raw($req, $git, $hex);
+		return git_tree_raw($self, $req, $git, $hex);
 	} else {
 		$type = 'application/octet-stream';
 	}
@@ -61,7 +61,7 @@ sub git_tree_sed ($) {
 # This should follow the cgit DOM structure in case anybody depends on it,
 # not using <pre> here as we don't expect people to actually view it much
 sub git_tree_raw {
-	my ($req, $git, $hex) = @_;
+	my ($self, $req, $git, $hex) = @_;
 
 	my @ex = @{$req->{extra}};
 	my $rel = $req->{relcmd};
@@ -83,10 +83,10 @@ sub git_tree_raw {
 	$qsp->psgi_return($env, undef, sub {
 		my ($r) = @_;
 		if (!defined $r) {
-			[ 500, [ 'Content-Type', 'text/plain' ], [ $git->err ]];
+			$self->rt(500, 'plain', $git->err);
 		} else {
 			$env->{'qspawn.filter'} = git_tree_sed($req);
-			[ 200, [ 'Content-Type', 'text/html' ] ];
+			$self->rt(200, 'html');
 		}
 	});
 }
diff --git a/lib/PublicInbox/RepoGitSrc.pm b/lib/PublicInbox/RepoGitSrc.pm
index 1546830f..de068940 100644
--- a/lib/PublicInbox/RepoGitSrc.pm
+++ b/lib/PublicInbox/RepoGitSrc.pm
@@ -30,9 +30,7 @@ sub call_git_src {
 			my ($info) = @_;
 			my ($hex, $type, $size) = @$info;
 			unless (defined $type) {
-				return $res->([404,
-					['Content-Type','text/plain'],
-					['Not Found']]);
+				$res->($self->rt(404, 'plain', 'Not Found'));
 			}
 			show_tree($self, $req, $res, $hex, $type, $size);
 		});
@@ -46,12 +44,12 @@ sub show_tree {
 	$req->{thtml} = $self->html_start($req, $title, $opts) . "\n";
 	if ($type eq 'tree') {
 		$opts->{noindex} = 1;
-		git_tree_show($req, $res, $hex);
+		git_tree_show($self, $req, $res, $hex);
 	} elsif ($type eq 'blob') {
-		git_blob_show($req, $res, $hex, $size);
+		git_blob_show($self, $req, $res, $hex, $size);
 	} else {
-		$res->([404, ['Content-Type', 'text/plain; charset=UTF-8'],
-			 ["Unrecognized type ($type) for $hex\n"]]);
+		$res->($self->rt(404, 'plain',
+			"Unrecognized type ($type) for $hex\n"));
 	}
 }
 
@@ -140,7 +138,7 @@ sub git_blob_sed ($$$) {
 }
 
 sub git_blob_show {
-	my ($req, $res, $hex, $size) = @_;
+	my ($self, $req, $res, $hex, $size) = @_;
 	my $sed = git_blob_sed($req, $hex, $size);
 	my $git = $req->{-repo}->{git};
 	if ($size <= $MAX_ASYNC) {
@@ -152,9 +150,7 @@ sub git_blob_show {
 			if ($ref eq 'SCALAR') {
 				$buf .= $$r;
 				if (bytes::length($buf) == $size) {
-					my $fh = $res->([200,
-						['Content-Type',
-						 'text/html; charset=UTF-8']]);
+					my $fh = $res->($self->rt(200, 'html'));
 					$fh->write($sed->($buf));
 					$fh->write($sed->(undef));
 					$fh->close;
@@ -163,13 +159,10 @@ sub git_blob_show {
 			}
 			my $cb = $res or return;
 			$res = undef;
-			$cb->([500,
-				['Content-Type', 'text/plain; charset=UTF-8'],
-				[ 'Error' ]]);
+			$cb->($self->rt(500, 'plain', "Error\n"));
 		});
 	} else {
-		$res->([200, ['Content-Type', 'text/plain; charset=UTF-8'],
-			[ 'Too big' ]]);
+		$res->($self->rt(200, 'plain', "Too big\n"));
 	}
 }
 
@@ -216,7 +209,7 @@ sub git_tree_sed ($) {
 }
 
 sub git_tree_show {
-	my ($req, $res, $hex) = @_;
+	my ($self, $req, $res, $hex) = @_;
 	my $git = $req->{-repo}->{git};
 	my $cmd = $git->cmd(qw(ls-tree -l -z), $git->abbrev, $hex);
 	my $rdr = { 2 => $git->err_begin };
@@ -239,9 +232,9 @@ sub git_tree_show {
 		my ($r) = @_;
 		if (defined $r) {
 			$env->{'qspawn.filter'} = git_tree_sed($req);
-			[ 200, [ 'Content-Type', 'text/html' ] ];
+			$self->rt(200, 'html');
 		} else {
-			[ 500, [ 'Content-Type', 'text/plain' ], [ $git->err ]];
+			$self->rt(500, 'plain', $git->err);
 		}
 	});
 }
diff --git a/lib/PublicInbox/RepoGitSummary.pm b/lib/PublicInbox/RepoGitSummary.pm
index 96ae9be9..ee9436f3 100644
--- a/lib/PublicInbox/RepoGitSummary.pm
+++ b/lib/PublicInbox/RepoGitSummary.pm
@@ -32,7 +32,7 @@ sub for_each_ref {
 	my $refs = $git->popen(qw(for-each-ref --sort=-creatordate),
 				EACH_REF_FMT, "--count=$count",
 				qw(refs/heads/ refs/tags/));
-	$fh = $res->([200, ['Content-Type', 'text/html; charset=UTF-8']]);
+	$fh = $res->($self->rt(200, 'html'));
 	# ref names are unpredictable in length and requires tables :<
 	$fh->write($self->html_start($req,
 				"$repo->{repo}: overview") .
diff --git a/lib/PublicInbox/RepoGitTag.pm b/lib/PublicInbox/RepoGitTag.pm
index 785de6b1..2a281ed6 100644
--- a/lib/PublicInbox/RepoGitTag.pm
+++ b/lib/PublicInbox/RepoGitTag.pm
@@ -78,13 +78,12 @@ sub git_tag_show {
 	my ($self, $req, $h, $res) = @_;
 	my $git = $req->{-repo}->{git};
 	my $fh;
-	my $hdr = ['Content-Type', 'text/html; charset=UTF-8'];
 
 	# yes, this could still theoretically show anything,
 	# but a tag could also point to anything:
 	$git->cat_file("refs/tags/$h", sub {
 		my ($cat, $left, $type, $hex) = @_;
-		$fh = $res->([200, $hdr]);
+		$fh = $res->($self->rt(200, 'html'));
 		$h = PublicInbox::Hval->utf8($h);
 		my $m = "git_show_${type}_as_tag";
 
@@ -97,7 +96,7 @@ sub git_tag_show {
 		}
 	});
 	unless ($fh) {
-		$fh = $res->([404, $hdr]);
+		$fh = $res->($self->rt(404, 'html'));
 		$fh->write(invalid_tag_start($req, $h));
 	}
 	$fh->write('</pre></body></html>');
diff --git a/t/repobrowse_git_log.t b/t/repobrowse_git_log.t
index 86338698..2caba546 100644
--- a/t/repobrowse_git_log.t
+++ b/t/repobrowse_git_log.t
@@ -10,7 +10,7 @@ test_psgi($test->{app}, sub {
 	my $req = 'http://example.com/test.git/log';
 	my $res = $cb->(GET($req));
 	is($res->code, 200, 'got 200');
-	is($res->header('Content-Type'), 'text/html',
+	is($res->header('Content-Type'), 'text/html; charset=UTF-8',
 		'got correct Content-Type');
 	my $body = dechunk($res);
 	like($body, qr!</html>!, 'valid HTML :)');
-- 
cgit v1.2.3-24-ge0c7